Phase identification using statistical analysis

1. A device comprising:

at least one processor configured to:

determine, based on (i) a first plurality of time-varying electrical measurements corresponding to a feeder head of a power distribution network having a plurality of phases and (ii) a second plurality of time-varying electrical measurements corresponding to a node in the power distribution network, and using statistical analysis, a predicted phase, from the plurality of phases, that corresponds to the node; and cause at least one device of the power distribution network to modify operation based at least on part on the predicted phase, wherein: using statistical analysis comprises determining at least one ranked correlation coefficient, the first plurality of time-varying electrical measurements comprises a respective set of time-varying electrical measurements for at least two phases in the plurality of phases, the second plurality of time-varying electrical measurements comprises time-varying electrical measurements for a single phase, and determining the predicted phase comprises:

determining, for each respective set of time-varying electrical measurements, a respective ranked correlation coefficient comparing the respective set of time-varying electrical measurements and the second plurality of time-varying electrical measurements; and determining, as the predicted phase, the phase corresponding to the respective set of time-varying electrical measurements having the respective ranked correlation coefficient with a highest absolute value.

2. The device of claim 1 , wherein the time-varying electrical measurements comprise voltage measurements.

3. The device of claim 1 , wherein determining the at least one ranked correlation coefficient comprises determining at least one Kendall rank correlation coefficient.

4. The device of claim 1 , wherein determining the predicted phase comprises performing principle component analysis to reduce a dimensionality of at least one of: the first plurality of time-varying electrical measurements or the second plurality of time-varying electrical measurements.

5. The device of claim 1 , wherein the time-varying electrical measurements comprise: active power consumption measurements, reactive power consumption measurements, frequency measurements, or current measurements.

6. The device of claim 1 , wherein causing the at least one device of the power distribution network to modify operation based at least in part on the predicted phase comprises outputting the predicted phase to another device.

7. The device of claim 1 , wherein causing the at least one device of the power distribution network to modify operation based at least in part on the predicted phase comprises: causing a node of the power distribution network to be switched to a different phase from the plurality of phases to improve load balance among the plurality of phases; or causing a smart inverter of the power distribution network to modify its output real power or output reactive power to regulate voltage on a phase from the plurality of phases.

8. A method comprising:

determining, by a computing device comprising at least one processor, based on (i) a first plurality of time-varying electrical measurements corresponding to a feeder head of a power distribution network having a plurality of phases and (ii) a second plurality of time-varying electrical measurements corresponding to a node in the power distribution network, and using statistical analysis, a predicted phase, from the plurality of phases, that corresponds to the node;

and

causing at least one device of the power distribution network to modify operation based at least on part on the predicted phase, wherein: using statistical analysis comprises determining at least one ranked correlation coefficient, the first plurality of time-varying electrical measurements comprises a respective set of time-varying electrical measurements for at least two phases in the plurality of phases, the second plurality of time-varying electrical measurements comprises time varying electrical measurements for a single phase, and determining the predicted phase comprises:

determining, for each respective set of time-varying electrical measurements, a respective ranked correlation coefficient comparing the respective set of time-varying electrical measurements and the second plurality of time-varying electrical measurements; and

determining, as the predicted phase, the phase corresponding to the respective set of time-varying electrical measurements having the respective ranked correlation coefficient with a highest absolute value that.

9. The method of claim 8 , wherein the time-varying electrical measurements comprise: voltage measurements, active power consumption measurements, reactive power consumption measurements, frequency measurements, or current measurements.

10. The method of claim 8 , wherein determining the at least one ranked correlation coefficient comprises determining at least one Kendall rank correlation coefficient.

11. The method of claim 8 , wherein determining the predicted phase comprises performing principle component analysis to reduce a dimensionality of at least one of:

the first plurality of time-varying electrical measurements or the second plurality of time-varying electrical measurements.

12. The method of claim 8 , wherein causing the at least one device of the power distribution network to modify operation based at least in part on the predicted phase comprises:

causing a node of the power distribution network to be switched to a different phase from the plurality of phases to improve load balance among the plurality of phases; or

causing a smart inverter of the power distribution network to modify its output real power or output reactive power to regulate voltage on a phase from the plurality of phases.